In my last post, I promised you I’d be taking a close look at the many factors that affect healer gameplay as an expansion progresses. And I got started on tackling the most theoretically challenging aspect – the throughput value of increased Spirit – right away. It seemed like a non sequitur, probably, but it was the cornerstone of an entire analysis on the influence of gear on healer throughput. In this post, I’ll be expanding on that analysis, to explore the magnitude of effect that stat growth has on healer throughput, and also to look forward to Legion with what little we know.
This will be a graph-heavy post, so if you are math-averse, I’m really sorry, but I’m not sure there’s any cleaner way to express the information.
I’ll just jump straight into the Big Picture, and drill down from there, highlighting a few interesting points along the way.
The Big Picture: Warlords Scaling
I turned to my heavily modded version of HealerCalcs to estimate the percent throughput increase that healers have experienced over the course of the expansion. I won’t bore you with the details of how I generated the data, other than to note that I used my first-week-of-raiding iLvl of 643 as the starting point, and the max iLvl recorded for healers on Warcraft Logs at the moment I’m making this post, 745, as the end point.
Here is the graph I generated, plotting throughput increase as a function of iLvl:
The green line at the top represents the combined effects of all of our gear on our potential maximum throughput. You can see that it’s just over 400%, which means that our HPS now at the end of HFC is potentially 5 times as strong as our HPS was at the start of Highmaul.
We turn now to our expert healing correspondent, Gwen Stefani, for her reaction:
And that pretty much sums up my opinion, as well. You’ve heard me rant here, or on the podcast with Hamlet, about how healing gameplay is most rich and exciting when we are weak compared to the incoming damage, and how it degenerates when we are strong. I just don’t think a 5X increase in our healing capacity (or over double the increase in player health) is fun, and I think that it leads to a lot of the problems we see at the end of an expansion – a reduced number of healer spots in raids, the spiky and spammy nature of healing, the loss of any consideration for efficiency.
The yellow line at the bottom is my estimation of Spirit’s contribution to throughput, taken from my previous post. And the purple line is showing the rest of our stats’ influence on our throughput: Intellect, Spellpower, and the 5 secondary stats (Critical Strike, Haste, Mastery, Multistrike, and Versatility). Those two lines, when multiplied together, create the green (total throughput) line.
The red line is there for reference: it is the increase in a player’s health pool as they gear up.
There are a lot of little intricacies hiding out in this graph, though, that might be worth exploring. And you know how I love exploring intricacies, so…
The Medium Picture: Stamina and Intellect
The green and red lines are the two most important to keep in mind for describing the way healing has felt during this expansion. While players’ health pools are still larger than any single heal a healer can cast (Selfless Healer/Libram of Vindication Paladins excepted), our capacity to heal a player up from damage rises at a much higher rate than that player’s health pool does. We’re not literally healing a target to full health in a single GCD (again, Selfless Healer/Libram of Vindication Paladins excepted – I need an acronym for this. SHLOVPE?), we are much more adept at handling the incoming damage than we were at the start of the expansion.
I’ve made this point before, but here is a great place to mention it again – player health acts as a “cap” of sorts on the amount of damage an enemy mob can deal with its attacks. As healers continue to gain power at a tremendous rate, the damage each attack deals can’t really go up at a similar rate. All the encounter devs can do is make the abilities deal that damage to more total targets, or more frequently, and in either case this leads to what I call “degenerate gameplay” of spam-and-AoE. So looking at what causes this discrepancy will help us understand why our gameplay always ends up deteriorating at the end of an expansion.
The main driver behind our power growth – and the reason it’s so huge compared to health pool growth – is the simple fact that healing scales with 7 stats. Intellect/Spellpower, Spirit, and all 5 secondaries contribute in some proportion to our total throughput. Meanwhile, player health rises with only a single stat – Stamina.
Since Intellect and Stamina are both primary stats, we should expect that they scale the same, and that Stam and Int cancel each other out. But I always like to check these things. I’m thorough like that.
As you can see, Int and Stamina track one another pretty closely. Int lags behind very slightly, because base Int (1130-ish) is a much higher proportion of our total Int than our base Stam (890) is of our total Stam. But it’s a very, very small difference. When we’re looking at this topic in the future, and want to isolate how healing outpaces health pool growth over an expansion, we can just leave Intellect/Spellpower out entirely, and consider only the secondary stats and Spirit as our scaling factors.
When we do this for the data presented in Figure 1, we get:
Over the course of Warlords, our healing grew 133%, or 2.33X, more than our raid’s health pool did. This is wholly attributable to the influence of Spirit and our secondary stats, and is the primary thing to measure when considering the negative gameplay effects of gear scaling.
The Detailed View: Secondary Stats and Spirit
But how much of the contribution comes from each source of power growth? Let’s drill down a little further.
In this graph, I’ve split out the Int/Spellpower contribution (blue line), and am comparing it here to the impact of Spirit, of all 5 secondaries combined (pink), the contribution of a favored secondary stat (green), and the contribution of an average secondary stat (orange). Some things to note:
- Spirit is a little over half as powerful as Intellect and Spellpower!
- All the secondaries combined are about 2/3rds as powerful as Spirit!
- If you stack a bunch of stat points into your favourite Secondary, let’s say it’s Haste here since we’re using a Druid profile at least nominally, it gains you like a 20% throughput increase over the course of the entire expansion. That sounds huge, but remember that our overall throughput actually went up by 400%, so your favourite secondary only explains about 5% of your total throughput increase.
- If you just shoved a random-ish amount of stat points into each Secondary, you get like 10% out of each secondary over the course of the expansion, or about 2.5% of your total throughput increase. The benefit of using the right secondaries is actually pretty low.
The first two points are the big takeaway here. Spirit is incredibly powerful, and it was the primary driver in our healing power’s growth relative to health pools. This was a problem we had all expansion as Spirit was mostly budgeted to trinkets, and trinkets were a bit mishandled. Secondaries themselves are a very, very small part of our overall throughput, and really aren’t the source of our problems.
The Side-Eye View: The Value of Optimizing Secondary Stats
Theorycrafters spend a lot of time and effort trying to determine which secondary stat is the best, and how to gear/gem/enchant. I have never spent much time working on stat weights as I personally don’t find the question to be that interesting, but when I was running these numbers and interpreting the data that led to me writing those last two dot points in the above list, I was very surprised by the findings. I guess I’d always assumed that using the right secondaries overall was a big difference in your throughput, and these numbers looked a little low to me, despite having agreed with Hamlet’s post about the overblown importance of the “right” secondary stats. Because I am perpetually distracted by shiny things, I thought I’d take a side trip to look a little closer at this phenomenon.
The very first graph in this post was generated using an essentially random smattering of secondary stats, and they ended up being pretty evenly distributed. This would be a fantastic plan if we assumed that every secondary stat affected our throughput by an identical amount – if Haste was equal to Mastery was equal to Crit, etc. (Compare 1.25 to 1.4 x 1.2 x 1.2 x 1.1 x 1.1; the former is slightly larger just by the way multiplication works.) But secondary stats don’t all affect our throughput by an identical amount, so I wanted to see how much of a difference it would make if I switched from a mostly-even distribution to a more optimized distribution.
So I reallocated my secondary stats from my gear profile to more resemble what a real Druid might do – 50% of available stats into Haste, 25% into Mastery, 12.5% each into Crit and Multistrike, and 0% into Versatility – and ran the numbers for Figure 1 again, expecting to see a fairly significant jump in my overall throughput from massively restructuring my gear:
I was pretty astonished. We went from a 403% increase with quite suboptimal stats to only a 407% increase – a whole 1% more throughput – from a pretty significant stat shuffle.
I mean … obviously for players who are min-maxing to get the most out of their character that they can, this is still a worthwhile pursuit. And for the rest of us, it’s – like Hamlet said, and as Talarian showed from a DPS point of view here – more of a messaging device than a throughput device. The right stat does improve your healing, by a tiny but not imperceptible amount, and it’s pretty easy to just gem and enchant for the right stats once you know what they are. But a lot of players – myself most definitely included – could get a lot more out of practicing their class (or improving their UI or reading a boss guide) than out of running to sims every time they get a new item and working out whether they should change their enchants.
The other conclusion from this is that, because Intellect is so much stronger than secondary stats are (about 3x as powerful as all the secondary stats combined, see Figure 5 above), iLvl is by far the most important statistic on our gear that we should be worried about. Secondaries can help us pick between two identical-iLvl items, but we probably shouldn’t be taking an item with a lower iLvl just because it has on it a preferred secondary stat. This was something we all suspected, but I was surprised by the magnitude of the effect.
The Long View: Legion Changes that Affect Scaling
Despite being in the Alpha test for Legion, I still don’t actually know a whole lot about what the expansion is going to look like when it goes live. But we have some tidbits about what is changing; here’s a list of the ones we know about that might affect these power growth issues, one way or another:
- Spirit has been removed from items, and mana regen is fixed at 4.0% max mana per 5 seconds (6400 mp5 at level 100, 44000 mp5 at level 110)
- Multistrike will no longer exist
- Intellect is removed from rings and perhaps necklaces (getting inconsistent results just browsing the gear that’s datamined on Wowhead)
- Raid-wide buffs will no longer exist. Some classes will be able to buff a limited number of players, but I’m not including these.
- Secondary stats may be allocated differently in Legion, so that secondary stats do not grow as much over the course of the expansion
- Legendary items may provide powerful effects that aren’t easily expressed in terms of stat growth
- Artifact Power, Artifact Traits, and Relics are all going to increase player power
Some of these effects are unknowable – we don’t have full details yet on points 5-7, for example – but others are quantifiable.
I’ve bolded the Spirit information because it is a big freaking deal and it’s not something that a lot of people seemed to have noticed yet. I do want to cover how mana will work in Legion without Spirit, but that’s going to have to wait for another post. For right now, it’s important to consider that the loss of Spirit is the loss of a very significant scaling factor for healers. Just looking back at Figure 1, our throughput increase will drop from that green line to the purple line just from that change alone.
The removal of Multistrike will also slightly reduce our scaling problems. It shouldn’t affect the total number of secondary stats we’re allocated, but now we’re only splitting our secondary stat points between 4 categories instead of 5. The biggest effect is just removing a multiplicative factor – 1.25 is larger than 1.254 – resulting in less snowbally behavior as we continue to accumulate gear. But also note that the loss of Multistrike means that it’s probably going to be hard to avoid having to take Versatility, and points into Versatility are typically worth less for us than points into the other stats. The overall effect of this will be quite small, since the contribution from secondary stats is already so small, but it will be another downward pull on our power growth.
Losing Intellect on rings (and necklaces?) has the potential to be a bit of a bigger deal, because Int is so strong for us, but rings and necklaces don’t really have a lot of stats on them anyway, so it’s a small change to a powerful stat. Just to check whether Int and Stam would still track in Legion, I quickly took the Int from rings and necklaces out of my original Figure 3:
It does obviously make a difference, but it’s not a huge one. If base Stamina and base Intellect stay proportionally similar in Legion as they are in Warlords, we shouldn’t see too much of a reduction in our scaling from this one change.
Losing raid buffs, however, should make the effects of the stats we do gain on our gear a little more prominent, just because we “start out” with less of it.
Legion Scaling Predictions
With these four mostly-knowable things in mind, I decided to re-run my original Healer Power Growth model after making some key Legion-based assumptions:
- Nothing is fundamentally changing about the way any of the stats work
- We will see another two-raid-tier expansion like Warlords, and will see our iLvl rise a similar amount to Warlords
- We’ll start at around 850 iLvl, and finish around 955 iLvl
- No sources of mana gain from gear
- No Multistrike
- No raid buffs
- No Intellect on rings or neckpieces
The result is a much flatter power curve over the course of the expansion:
An overall growth of 293%, or about 4X throughput, is still pretty high. Though if we continue to assume that Stamina and Intellect/Spellpower will track similarly over the course of Legion, then the pink line describes how our healing will increase relative to our raid’s health pool – a much smaller 60% increase (or 1.6X throughput) to our healing. That is a big improvement.
But it’s a larger secondary-stat contribution than we had in Warlords! This is probably from losing the raid buffs and because, if the secondary stat rating-to-percentage conversions I found for level 110 characters are accurate, our ratings are taking less of a hit, so we’re starting out with a larger amount of secondaries, and gaining a lot more than we gained in Warlords as we gear up.
I’ll note that Celestalon’s post about secondary scaling had the following benchmark in it:
The % of a tank’s healthbar that you can heal per second went up over the course of Warlords, by a massive amount (probably about 3x). We don’t want it to not go up (it gives a good feeling of progression), but it should only go up a little (probably to 1.5x).
That 3X idea for Warlords was a bit high – it was really more like 2.3X from the combination of Spirit and secondary stats (see Figure 4). But we’re pretty close to his new benchmark in Legion already with this 1.6X increase from secondaries without even changing the way they scale.
However, we’re still overshooting it, and having a 1.6X increase from secondaries alone leaves us no room for things like introducing crazy powerful trinkets/Legendaries, Artifact Traits, or even for reintroducing some limited mana regen on gear or from tier set bonuses. So just for funsies, I thought I’d look at what might happen if we halved the amount of secondaries we get per iLvl, and see how our scaling would look. It’s a complete spitball by me – I have no idea what they’re really going to do with secondary stats – but I just wanted to see how drastic of an effect this might have:
As a result of halving our secondary stat acquisition, we go down to a 3X increase in healing power over the course of the expansion, and only a 22% increase in healing from secondaries alone. This leaves ample room for Legendaries, Artifacts, trinkets, and set bonuses to boost our power without overshooting Celestalon’s target.
Combining multiplicative factors with exponential scaling results in undesirable power growth over the course of an expansion. One obvious solution would seem to be, “just go linear with stat gains”, but I think the main reason we don’t see that happening – besides whatever technical issues it might pose, with WoW being 10 years old and having nearly 1000 iLvls of gear – is that linear growth would feel vastly different. There’d be a very real concern that linear growth might mean that players don’t feel compelled to attempt the next tier of raiding because its item rewards aren’t worth the additional difficulty. (I mean, we reportedly saw this behavior in Warlords with exponential gear, which is why we saw a 5-iLvl bump to Blackrock Foundry’s gear rewards a couple weeks after its release. So it’s not infeasible that linear growth would also incur this same problem.)
Also, I think there may be some value in letting players get more stronger during the final tier than during the initial tier, if the final tier is going to last so long – it’s at least a source of amusement. But really, a linear model and a shorter final tier would probably be just fine, at least from my perspective. Hamlet and I discussed the idea of “going linear” for at least one stat – Spirit – on our podcast, and this could have even been an alternative to eliminating the stat entirely.
Changing to a linear growth model is not really necessary, though. The big contributor to our inflated healing power this expansion was Spirit. While not every class valued Spirit at the start of Warlords, the changes to Mistweaver Monks and Holy Paladins encouraged those classes to become Spirit-lovers, and I think perhaps the Disc Priest’s Repudiation of War was meant to serve a similar function (but if so, I think it has failed). So even if a HPal or Monk’s throughput-from-Spirit curve started out very, very flat, after these changes it probably followed a similar trajectory to the traditionally Spirit-loving Druid. With Spirit being gone in Legion, we’re mostly on our way to having a more sane growth model than we’ve seen in a long while.
I know that it seems overly optimistic to end a blog post with “Legion will fix it” but in this case, you guys, I think Legion really will fix it. There’s still a lot of moving parts we don’t understand, about Artifacts, Legendaries, set bonuses and trinkets, that have the potential to skew our power growth for the worse, but depending on how the secondary scaling changes are implemented, the devs may have a lot of room to work with to still give us powerful items without breaking our game. At the very least, I think we can see that this issue is on the devs’ minds, that they recognize it’s a problem, and that they want to fix it. That’s all I could really ask for!